Does la_solve not attempt least-squares solutions?

According to the LinearAlgebra documentation for la_solve:

"

If the matrix is not square, we return a least-squares solution computed by

performing a QR factorization of the matrix.


If the number of rows of the matrix does not match the number of rows of

the right hand side object, the returned object has status

LA_DIMENSION_MISMATCH_ERROR.

"

However, my tests indicate that la_solve returns LA_DIMENSION_MISMATCH_ERROR, including when set up for a least-squares solution (i.e. number of rows of matrix and vector match). Please provide working sample code for a least squares computation. Preferably in Swift 3.

Accelerate > vecLib > LinearAlgebra

https://developer.apple.com/videos/play/wwdc2014/703/

Only describes the use of la_solve with square matrics. Which documentation is correct? Video or headers?

Is there any documentation beyond WWDC2014/703 and the minimal entries in Xcode? I've been unable to find anything on the la_xxx commands, yet there are significant differences between the BLAS commands and la_xxxx commands like la_solve.

Does la_solve not attempt least-squares solutions?
 
 
Q